You see how it’s going, without asking.
Goals measure themselves from finished work. Open the board and the answer is already there.
We don’t put AI on top of your processes: automating one that doesn’t work only makes it fail faster. In six weeks we look at how your team works with you, rebuild one process so your people and their AI agents share the work, and Oruscribe records every goal, decision and task to show you the before and after.
Already have an account? Sign in →
Compatible with the AI assistants that use MCP
- Claude CodeVerified end to end
We check the other MCP-compatible assistants with you when the pilot starts.
01 — The problem
Faster, not richer.
What it’s costing you
- 01
You can’t show what it produced
The team delivers faster, but the proof of what was delivered lives in chats, commits and each person’s memory.
- 02
Every agent works alone
Two people, or a person and their agent, solve the same thing without knowing. What was decided never reaches the next agent.
- 03
Reviewing became the bottleneck
Agents produce more than a person can review.
- 04
You sell hours in a world that shrinks them
If AI saves you hours and you bill by the hour, you earn less. To charge for results you have to be able to prove them.
The context
- 42%
of Colombian companies already use AI. Among those surveyed, only 29% measure business results and almost half focus mainly on activity metrics: users, queries processed, time saved.
AWS study by Strand Partners, “Desbloqueando el potencial de la IA en Colombia” (“Unlocking the potential of AI in Colombia”; 2026; 1,000 business leaders), as reported by Portafolio and Infobae. A vendor’s study.
- ~6%
of the companies McKinsey surveyed attribute at least 5% of their operating profit to AI and report significant value. Those that do were almost three times as likely to have redesigned their workflows.
McKinsey, The State of AI in 2025 (global survey).
- +98%
more merged PRs on teams with heavy AI use, but with +91% review time and no improvement observed at the company level.
Faros AI, The AI Productivity Paradox Report 2025 (July 2025; 10,000+ developers). A vendor’s analysis, correlational.
Using AI is not the same as capturing its value.
02 — How it works
From goal to result, with proof at every step.
Oruscribe records every link and ties it to the next. The first ones already work; we build the rest with the first pilots.
- 01
Goal
What you want to achieve and by when. Progress is calculated on its own, from finished work.
Available - 02
Decision
What was decided, why, and what was discarded. If it changes, the history stays.
Available - 03
Task
Who does it, person or agent, and how it’s going. If it moves to someone else, the credit for earlier work is kept.
Available - 04
Deliverable
What was delivered, attached to the task and checked against its source; first, GitHub PRs.
In development - 05
Review
Who approved it and by what criteria. Low-risk work passes on its own; what reaches the client is approved by a person.
In development - 06
Result
How much the agreed metric improved, measured against a baseline taken before the change.
In the pilot
- Availableit works today.
- In developmentwe’re building it with the first pilots.
- In the pilotwe do it with you, with hands-on support.
03 — What already works
What changes on Monday morning.
You open one board and know how everything is going: your people, their AI agents, the goals and whatever got stuck.
On trackAt riskBlocked
You know who needs help.
Each person and each agent reports how it’s going. You see the risk before it reaches the client.
BlockerTask stuck
MilestoneGoal at 80%
InfoNew member
If something gets stuck, you find out.
A blocker raises an alert for the team lead, without waiting for Friday’s meeting.
Client store on WooCommercesuperseded
Migrate the client store to Shopifycurrent
Nothing gets lost.
Every decision is recorded with its why. If it changes, the full history stays.
AL MR AGENT
You know what your AI is doing.
Each agent is a team member with its own profile and history. You see what it did, when, and on whose behalf.
LIVE
Always up to date.
The board updates itself and every assistant checks the latest when it starts. Nobody works from old information.
Illustrations of how Oruscribe looks; the data is sample data.
04 — Continuity
Knowledge stays with the company. Credit stays with whoever did the work.
No one should be the only person who knows how something works. Oruscribe keeps the work on record as it happens, so every handover is calm: for the company and for the person handing over.
The work carries on.
Vacations, promotions, team changes or departures: whoever takes over sees what was decided, why, and where each task stands.
Your work stays recognized.
Every task keeps who did it, even if it later moves to someone else. You can rest, grow or change areas without everything depending on you.
-
Decides to renew with the supplier and records why.
Decision -
Works the task and leaves it in progress.
In progress -
Changes area and hands the task over with a note.
Handover -
Picks it up with all the context and finishes it.
Done
Credit for the earlier work: AL
05 — Ask
Ask your company.
Your usual assistant reads the record of the work —goals, tasks, decisions and statuses— and answers with real data, remembers every decision and helps you see what makes sense to do. The final call is yours.
Illustration
You
How is the team doing this week?
Your assistant
- At risk.
- 1 active blocker: waiting for API access, since yesterday.
- Sales, the most delayed goal, is at 31%.
- 5 tasks were finished this week.
You
Why did we decide to migrate the store to Shopify?
Your assistant
- It was decided three weeks ago.
- Reason: the client needed local payments and a faster checkout.
- It replaced the decision to keep WooCommerce, which remains in the history.
You
What should I prioritize this month?
Your assistant
- Sales is your most delayed goal: 31%.
- API access is blocking part of that work.
- Unblocking it moves it the most.
- You decide how to proceed.
And if you’d rather see it on a board, it’s the same thing, in one view.
To do3
Define Q4 scope
AL
Draft proposal
MR
Review contract
JD
In progress2
Integrate invoicing
JD
Migrate board
AG
Blocked1
Wait for API access
AL
Done3
Team onboarding
JD
Agent playbook
AG
Weekly report
AL
06 — Why Oruscribe
What makes Oruscribe different.
We rebuild the process, not just automate it.
Automating a process that doesn’t work only makes it fail faster. In the pilot we look at how the work is shared with you and rebuild one process designed for people and AI agents working together.
In the pilot
Works with the AI you already use.
Assistants that connect through MCP work on a single source of truth, even if your team uses several brands.
Available
Keeps the why.
Decisions are objects of their own, with their reasoning and what was discarded. Assistants can look them up before proposing something that was already discarded.
Available
Evidence, not testimony.
Instead of relying on the agent to tell you what it did, the deliverable is attached to the task and checked against its source. GitHub PRs first; with Claude Code, the PRs the agent creates could be attached automatically if each person turns it on.
In development
Review at scale.
Explicit acceptance criteria, risk levels and a fast review queue. Every approval is recorded.
In development
Measured results, in Spanish.
Goals measure themselves from finished work. The before and after of the redesigned process, with a report you can show your client, we do with you in the pilot. All in Spanish, with direct conversation on WhatsApp.
AvailableIn the pilot
07 — Compared to other options
Why isn’t what I already have enough?
What teams that already use AI ask us most.
Why not just pay for Claude, ChatGPT or the assistant we already use?
Keep using it: Oruscribe connects to it, it doesn’t replace it. An assistant is a very capable worker, but it doesn’t sit down with your team to look at how the work is shared, redesign a process or measure whether it improved: we do that in the pilot. And Oruscribe is the common place where everyone’s work stays, people and agents, with its why, without depending on one person’s account or on a single assistant brand.
I already use Claude every day. What do I get from Oruscribe?
Your Claude stops knowing only what you tell it. Today it doesn’t know what your teammate decided yesterday, why the other option was discarded or which task was left half done. With Oruscribe connected, which with Claude Code is one command, your Claude can look up the team’s goals, decisions and tasks before proposing something. It doesn’t compete with Claude: it gives it the team’s memory.
What if I connect Asana, ClickUp or Trello to my assistant through MCP?
You can, and it works for tracking task status. But a task tool records what needs doing and whether it got done. Oruscribe also keeps why each thing was decided and what was discarded, who did each task, person or agent, and how each goal is progressing: what whoever comes next needs to pick up without asking. And it’s not just software: it comes with the process redesign and the before-and-after measurement. For the pilot we don’t ask you to drop your tools.
Isn’t this the same as hiring an automation?
No. Automating puts AI on top of the process you already have; if that process has waiting, handoffs or rework, it only makes them faster. We first look at how your team works, rebuild one process so people and agents share the work differently, and measure whether it improved. In the pilot we don’t build custom automations: we change how the work is done and leave it measured.
08 — Who it’s for
Made for agencies and digital service teams.
Oruscribe works where several people and AI agents need to coordinate, and real teams already use it.
Development agency on a retainer
Every delivery is tied to its goal, its decision and its PR. When the client asks what they received, you open the record instead of reconstructing the story.
Checked PR: in development
Agency with several accounts
One team per client, each with its people and its agents. You see the status of every account without asking for a single report.
Available
Team that wants to sell results, not hours
You agree a metric with your client, we measure it before and after, and the report backs the pricing conversation.
In the pilot, with support
Illustrative scenarios of how Oruscribe is used.
09 — The pilot
Six weeks, one process, a before and an after.
A pilot with hands-on support: we work with you on one concrete process and measure the result.
- Week 1
Diagnosis
We look at how your team works today: handovers, waiting, rework. We pick a metric with you and take its baseline.
- Weeks 2 and 3
Redesign of one process
We redesign one process around a person and their agents. You don’t change tools; how the work is shared changes.
- From week 2
Oruscribe installed
We connect your team and their assistants; goals, decisions and tasks get recorded as the work happens.
- Weeks 4 to 6
Measurement
We measure the agreed metric and hand you the before and the after.
- Afterwards
If you decide to continue
Oruscribe moves to a monthly subscription, with no lock-in.
What you get at the end
- A redesigned process, up and running.
- The record of the pilot’s work: goals, decisions, tasks and statuses.
- A report of the before and after of the agreed metric.
Price for the first teams
USD 2,400for the six weeks
For a team of up to 10 people and their AI agents. Includes the diagnosis, the redesign of one process, Oruscribe installed and the before-and-after report. Taxes not included.
Part tied to improvement
30% of the price (USD 720) is charged only if the metric we agree in week 1 reaches the agreed target. If it doesn’t, that part isn’t charged and you still get the report.
After the pilot
Monthly subscription of USD 15 per person (minimum 5 people), with your AI agents included and no lock-in.
10 — Control
Everyone sees what applies to them.
Oruscribe respects how your company is organized: leadership sees the big picture of the work and each person sees their own. An AI agent can be one more team member, with its own profile and permissions, or work as one person’s assistant, with that person’s permissions: no more, no less.
Manager · Coordinator · WorkerAI agent
Manager
Sees the whole organization: teams, goals and the state of the work, whether it’s done by people or by agents.
Coordinator
Manages their team: assigns tasks, sends alerts and sees how their people’s work is going.
Worker
Manages their work, records decisions and reports how it’s going. Can be a person or an AI agent.
11 — Security
Your information, under your control.
- 01
Every company, isolated
Each company’s data lives apart from everyone else’s, with rules enforced in the database itself.
- 02
Permissions by role
Each person and each agent sees only what their role allows.
- 03
Everything on record
Tasks, decisions and status changes keep their history. Nothing is deleted: it’s archived.
- 04
Protected keys
For each access key we only store a fingerprint, never the key itself. And you can revoke it whenever you want.
3 permission levels
1 place for people and agents
256 bits of fingerprint protect every key
12 — How it connects
Nobody switches tools. Everyone gets connected.
- 01
We set up your organization with you
We prepare your space, you invite your people and, if you want, add your AI agents as one more member.
- 02
Each person connects their assistant
With Oruscribe’s URL and by signing in with their account, no tokens. With Claude Code it’s one command; we check the other MCP-compatible assistants with you.
- 03
Everyone works on the same thing
Tasks, goals and decisions in one place. Each role sees what applies to them.
For whoever sets up the connection
Oruscribe uses MCP, the open standard AI assistants use to connect to other tools. You don’t need to understand it: this is what whoever sets it up will do with Claude Code.
claude mcp add --scope user --transport http oruscribe \
https://cfamprdygpngrjtaxzes.supabase.co/functions/v1/mcp-server
# then, inside Claude Code: /mcp → oruscribe → Authenticate13 — FAQ
What people ask before getting started.
What does the pilot include?
A diagnosis, the redesign of one process, Oruscribe installed and the measurement of the before and after, over six weeks. Service is included: we work with you, we don’t just give you access to software.
How much does the pilot cost?
USD 2,400 for the six weeks, for a team of up to 10 people and their AI agents; if your team is larger, write to us and we’ll adjust it. Of that price, 30% (USD 720) is charged only if the metric we agree in writing in week 1 reaches the agreed target, as long as the team has used the redesigned process; if it doesn’t, that part isn’t charged, you still get the report and you don’t have to continue. If you decide to go on, the monthly subscription is USD 15 per person (minimum 5 people), with your AI agents included and no lock-in. These are prices for the first teams and don’t include taxes.
Does my team have to switch tools?
No. Each person keeps using the AI assistant they already use. Oruscribe connects to those assistants and gives them a common place to work.
Do I need a technical team to get started?
Not to use it. We set up your organization with you and help you connect the first person and the first agent. After that, each person connects theirs in a few steps.
What can I ask Oruscribe?
Anything about the work of your team and its agents: how each goal is going, what got stuck, who did what and why each decision was made. The answer comes from the assistant you already use, with what Oruscribe knows.
How do I explain this to my team?
With transparency. Oruscribe doesn’t capture screens, keyboards, commands or time: it stores what gets reported —tasks, decisions and statuses— and, when it’s available, the deliverables (a PR, for example) attached to a task; each person decides whether to turn on automatic attaching. Every task keeps who did it, even if it moves to someone else: the record is there to give credit, not to watch people. And each person sees their own work.
How secure is our information?
Each company’s data is kept isolated, each person and each agent sees only what their role allows, and everything keeps its history. We only store a fingerprint of access keys, and you can revoke them whenever you want.
What if an agent gets it wrong?
Every action is attributed: you know what happened, when, and on whose behalf. And since nothing is deleted, you can correct it and the full history stays.
What kind of company is it for?
For agencies and digital service teams that need to show what they produce, starting with development agencies. If your people already use AI assistants every day and your clients ask what they received, it’s for you. If you don’t work with AI yet, write to us anyway: starting from scratch is a different scope and we’ll talk it through with you.
We’re adding teams gradually. Message us on WhatsApp with your company, your team size and the AI tools you use, and we’ll reply to see whether the pilot makes sense.
The conversation starts on WhatsApp. No forms.
Already have an account? Sign in →